Blog

We wanted convenience. Companies gave us automation. Somewhere along the way, reaching another person became the hardest part of customer service.

Over the past couple of months, I found myself needing help from several large companies. The businesses were completely different and, initially, I didn’t think much of the incidents because things go wrong. A shipment doesn’t arrive, a service doesn’t work as expected, a payment or refund gets delayed, a phone is stolen and needs to be blocked. These are all things that can happen, even with established companies.

What stayed with me was not so much that something had gone wrong, but what happened when I tried to get it sorted.

In one case, through my company’s account with a delivery aggregator, we shipped 20 prepaid packages. None of them were delivered. The courier partner initially refused the shipments and then asked us to change the envelopes. We did that, but the shipments were refused again. At the same time, the tracking information did not appear to reflect what was actually happening, and the tickets raised through the corporate account did not give us a clear resolution. I kept trying to find someone who could look at the situation as a whole and simply explain what was happening and what could be done.

About a month later, I had a completely different problem when my phone was stolen. I contacted my mobile operator to have it blocked. There were tickets raised through the corporate account, conversations at the store, emails and repeated attempts to get the matter addressed. At the store, I was told that the staff could forward the required documents but could not actually resolve the issue there, while the responses I received by email were largely standard replies. Again, what I was looking for was not particularly complicated: I wanted someone to understand the situation, tell me what needed to be done and help me get it done.

The phone was eventually blocked. The SIM replacement, however, happened only after I raised the issue publicly on LinkedIn.

There were other, smaller experiences around the same period as well: an online order that was not delivered, another that arrived late, and a return that took weeks to be refunded. Each of these, on its own, is hardly unusual. Most of us have experienced something similar, and I certainly don’t expect every company to operate without mistakes.

What I do expect is that when something goes wrong, there should be a reasonably straightforward way to get help.

That sounds obvious, but increasingly, it isn’t.

Most companies today have invested heavily in making customer service available through technology. There are apps, help centres, automated emails, chatbots, AI assistants, automated phone menus and self-service systems that can respond at any hour. In many situations, this is genuinely useful. If I want to track a parcel, download an invoice, reset a password or check the status of an order, I would much rather get an immediate answer than spend fifteen minutes waiting for an employee to tell me something the system already knows.

So this isn’t an argument against AI or automation. There is a lot that technology can do very well, and there is no reason to make a human handle a simple request when a machine can resolve it accurately and immediately.

The difficulty starts when the problem is not straightforward, or when the automated system simply cannot deal with what has happened. It becomes even more frustrating when, after trying the automated route, I explicitly say that I want to speak to an agent.

At that point, I have already made my choice. If I say, “I want to speak to an agent,” why should the system continue trying to convince me that it can solve the problem?

I understand the company’s side of this. Automation is faster, scalable and cheaper, and no business can realistically have a human employee available to handle every routine customer interaction, twenty-four hours a day. If a chatbot can answer a simple question in seconds, it makes sense for the company to encourage customers to use it, and in many cases it makes sense for the customer as well.

But there is an important difference between encouraging self-service and forcing self-service. If I want to use the chatbot, I’ll use it. If I want to solve the problem myself, give me the tools and I will. If the automated system can solve my problem, that’s probably the easiest outcome for everyone. But if I have already tried that route and I specifically ask to speak to a person, and human support is available, the company should connect me.

The company can tell me it believes it can resolve the issue, or offer me one more thing to try first. But if I still want to speak to an agent, that decision should be respected.

Instead, what often happens is that the customer is taken through another set of questions, another menu, another suggested article or another automated response. Eventually, the effort required to get through the customer-service system becomes greater than the effort required to deal with the original problem.

And that is where I think we need to pause and ask ourselves what we actually mean by customer service, because when everything works normally, almost any system can look efficient. The real test comes when something doesn’t work as expected, when the customer’s situation doesn’t fit neatly into a predefined category and when the answer isn’t sitting somewhere in a help centre.

That is when I don’t necessarily need more information. I need someone to understand what happened and help me work out what can be done. There is a difference between those two things, and it is a difference that automation has made easy to overlook.

Information and resolution are not the same. Automated systems are very good at providing information. They can tell me what the policy says. They can point me to a page that might contain an answer. But giving me information is not the same as taking responsibility for a situation, and it is not the same as resolving it.

I think we also underestimate how much of customer service is simply about being heard. Sometimes the problem gets solved in that first conversation. But even when it does not, there is a meaningful difference between being told “according to our policy, this cannot be done” and hearing “I understand why this is frustrating, let me see what options we have.” The final outcome may be exactly the same. But the second experience leaves you feeling completely different, because someone listened, someone acknowledged that there was a person on the other side of the transaction, and someone took ownership.

I am not arguing that human beings are always good at this. We have all dealt with agents who read from a script, transfer us to another department, or tell us it is not their problem. A human being is not automatically good customer service.

So the real question is not whether companies should choose technology or people. It is how the two should work together.

The sensible design is fairly straightforward. Let the technology handle what it is genuinely good at: verifying identity, understanding the basic issue, collecting the relevant documents, checking previous interactions and summarising the problem. Then, when a human takes over, that person already knows what has happened. The customer does not have to start the story from the beginning. The technology is not replacing the human; it is making the human more effective. That is what a meaningful use of AI should look like.

Companies measure what they can count. They can tell us how many calls were automated, how many tickets were resolved and how quickly responses were sent. Those numbers matter, and it is reasonable to track them. But they do not capture how many customers simply gave up, how many had to repeat their story, or how many left the interaction resolved in the system’s terms but genuinely dissatisfied.

Efficiency is not the same as experience. If automation saves me five minutes, that is a win. If it saves the company money and costs me forty minutes of trying to reach someone, calling it an improvement requires some justification.

There is one detail from these experiences that I keep coming back to. In the two most serious cases, the situation moved forward only after I raised the problems publicly. The phone was eventually blocked, and the SIM replacement followed the LinkedIn post. The delivery issue also received attention after I raised it publicly. Before that, I had been trying to resolve both matters through official channels for some time.

I cannot know exactly why the response changed once the problems became public, and I don’t want to speculate about the reasons. But as a customer, it was difficult not to notice the difference. It also raises a question the companies have not answered: if a customer follows the official process, raises tickets and repeatedly asks for help, why should going public seem like a more effective route to getting attention?

None of this means companies should abandon automation, and it does not mean every problem deserves a phone call. Customers have contributed to this environment as well. We have demanded the lowest prices, instant service and 24/7 availability, and companies have responded in the only way that makes economic sense. Convenience, however, should not mean giving up the ability to reach a person when one is genuinely needed.

Perhaps the simplest rule is:

There will be exceptions. Agents may genuinely be unavailable. Some services may operate entirely through digital channels. There may be security or regulatory reasons why an interaction has to follow a particular process. Those exceptions should be stated clearly rather than hidden behind menus. Companies should not pretend there is a human waiting behind the next button when there isn’t, and they should not make customers feel that asking for a person is an unreasonable demand.

Beyond convenience, this is about trust. When I buy something from a company, I am not buying from an algorithm. I am buying from a business made up of people, and when something genuinely goes wrong, I expect that business to stand behind what it sold. That does not always mean giving me exactly what I want. It means giving me a fair opportunity to explain what happened and having someone consider it.

Perhaps that is the standard worth protecting as technology becomes more embedded in customer service. Not human beings everywhere. Not AI everywhere. But the right kind of help at the right time.

The machines can handle the routine. The people can handle the complicated. And the customer should be able to move between the two without feeling trapped.

Because ultimately, the purpose of customer service should be simple: to help the customer when the customer needs help.

When I say, “I want to speak to an agent,” the most customer-friendly response is not another question.

It is simply:

“Of course. Let me connect you.”